An overview on reproductive isolation in Oryza sativa complex

Abstract Reproductive isolation is generally regarded as the essence of the speciation process. Studying closely related species is convenient for understanding the genetic basis of reproductive isolation. Therefore, the present review is restricted to the species and subspecies of the Oryza sativa complex, which includes the two domestic rice cultivars and six wild species. Although closely related, these rice species are separated from each other by a range reproductive barriers. This review presents a comprehensive understanding of the forces that shaped the formation of reproductive barriers among and between the species of the O. sativa complex. We suggest the possibility that domestication and artificial breeding in these rice species can lead to the early stages of speciation. Understanding the evolutionary and molecular mechanisms underlying reproductive isolation in rice will increase our knowledge in speciation and would also offer practical significance for the implementation of crop improvement strategies.


Introduction
Genetic divergence, reproductive isolation, natural selection and human-assisted artificial speciation are the vital forces that shape population genetics and consequent speciation (Liu et al. 2015;Schulter and Panell 2017). Reproductive isolation is a very important evolutionary phenomena that prevents genome homogenizations and maintains the integrity of species (Bomblies and Weigel 2007;Ouyang and Zhang 2013;Chen et al. 2016). The evolution of reproductive isolation allows differentiation and local adaptations to become fixed in diverging populations.
Rice belongs to the genus Oryza, which contains 25 recognized species, of which 23 are wild and 2 are domesticated (Vaughan et al. 2003). The genus Oryza has been classified into different species complexes based on their nine distinct genomes, viz., A, B, C, D, E, F, G, H and J (Vaughan et al. 2003). The O. sativa complex belongs to the A genome and contains two domesticated species O. sativa and O. glaberrima and six wild species: O. rufipogon, O. nivara, O. barthii, O. longistaminata, O. meridionalis and O. glumaepatula. These species constitute the primary gene pool of rice (Vaughan et al. 2003;Tripathi et al. 2011). The wild rice O. rufipogon is a perennial, outcrossing species widely distributed in Asia and Oceania. Oryza nivara is an annual, self-fertilized wild rice species mainly found in South and Southeast Asia. Oryza rufipogon and O. nivara are sometimes considered to be separate species or ecotypes of the same species (Sang and Ge 2007;Vaughan et al. 2008). The perennial O. longistaminata and the annual O. barthii (also called O. breviligulata) are the African wild rice species and can be found growing in the same area. Oryza longistaminata is a rhizomatous and self-incompatible species found to be the most diverged of all the species in the O. sativa complex (Vaughan et al. 2008). The annual O. glumaepatula is widespread in Tropical America, whereas the annual, inbreeding and highly diverged O. meridionalis is endemic to Tropical Australia and is often sympatric with Oryza australiensis (EE genome) in Australia. Oryza sativa is the Asian cultivated rice and is distributed globally, whereas O. glaberrima is the African cultivated rice and is mostly confined to Africa and differs from O. sativa in its morphology and ecology. Oryza sativa has been further subdivided into multiple varietal groups, the major ones being indica and japonica (Garris et al. 2005;Sweeney et al. 2007). In addition, weedy rice (Oryza sativa f. spontanea), which is conspecific and congeneric to cultivated rice, occurs together with cultivated rice in and around the rice fields (Nadir et al. 2017). The weedy rice associated with O. sativa may be called O. sativa, although they are not the crop. Those associated with O. glaberrima were sometimes called O. stapfii (Suh 2008).
The earliest evidence for the domestication of Asian rice, O. sativa found to date was at the region of the Yangtze River valley of China dated back to 11000-12000 BC (Vaughan et al. 2008;Gross and Zhao 2014). The wild Oryza species, O. rufipogon or O. nivara or possibly both of them, are the progenitors of O. sativa ( Fig. 1) (Sang and Ge 2007;Vaughan et al. 2008). Apparently, conflicting data are available supporting single and multiple events leading to domestication of O. sativa (Oka 1988;Cheng et al. 2003;Londo et al. 2006;Vaughan et al. 2008;Molina et al. 2011;Huang et al. 2012). Molecular studies based on similarity in the alleles for non-shattering grains and erect growth in indica and japonica subspecies support the hypothesis of single domestication event Lin et al. 2007;Jin et al. 2008;Tan et al. 2008). These studies suggest that after domestication O. sativa spread and diversified to create divergent subgroups. Other studies based on the biochemical traits, hybrid sterility and subsequently supported by molecular analyses (Cheng et al. 2003;Tripathi et al. 2011) suggest that indica and japonica subspecies originated under separate domestication events from two divergent wild rice species in China and India, respectively (Tripathi et al. 2011). The African rice O. glaberrima was domesticated from O. barthii separately but parallel to the Asian rice in the African continent between 300 BC and 200 BC during a single domestication event ( Fig. 1) (Murray 2004;Purugganan 2014). Oryza barthii was introduced from Asia into Africa (Vaughan et al. 2008).
Considerable high genetic variations exist in this primary gene pool of rice. For example, three regional variants of O. glumaepatula, five distinct groups of O. longistaminata and at least two different genetic groups in O. rufipogon, based on their ecology and life histories, have been recognized (Oka 1988;Akimoto et al. 1998;Vaughan et al. 2008). Similarly, the variation between indica and japonica subspecies of O. sativa is well documented (Garris et al. 2005). Differences were also recognized between temperate and tropical japonica varieties as well as within the tropical japonica varieties (Garris et al. 2005;Vaughan et al. 2008). Compared to O. sativa, O. glaberrima has a restricted geographic distribution and consequently lower genetic diversity exists in African rice (Semon et al. 2005). Only a few genetic subgroups have been detected in O. glaberrima, which reflects the ecological differentiation of O. glaberrima in different habitats (Semon et al. 2005).
Almost all kinds of reproductive barriers so far reported in plants have been found in O. sativa complex and these include reduced cross-fertility, low germinability of F 1 seeds, F 1 inviability, F 1 pollen and embryo sac sterility, and a sporophytic sterility and weakness in F 2 generations also known as hybrid breakdown ( Fig. 2) (Amemiya and Akemine 1963;Oka 1969, 1970;Ichtiani et al. 2007Ichtiani et al. , 2011Chen et al. 2008Chen et al. , 2014. Despite the occurrence of many barriers to hybridization, introgression among these Oryza species is common. Gene flow between O. sativa and O. rufipogon was identified in a number of studies (Pusadee et al. 2016;Wang et al. 2017). Gene flow has also been reported between O. rufipogon and O. nivara as well as between the indica and japonica subspecies of O. sativa (Zheng and Ge 2010;Yang et al. 2012). Hybridization between O. sativa and African wild and cultivated rice at varying level has also been reported (Jones et al. 1997).
Previously, comprehensive literature has been presented on the reproductive isolation in model as well as non-model organisms which has extending our understanding of reproductive isolation (Coyne and Orr 2004;Bombiles and Weigel 2007;Lowry et al. 2008;Ouyang and Zhang 2013;Baack et al. 2015). The topic of reproductive isolation in rice (O. sativa) has been reviewed in previous studies (Ouyang et al. 2010;Ouyang and Zhang 2013) but these studies have focused only on the hybrid sterility observed in indica-japonica hybridization. As rice is one of the better developed systems for understanding the evolution of reproductive isolation, we suggest that a broader view of the various factors that cause the reproductive isolation offers the opportunity to thoroughly understand the phenomena of reproductive isolation in rice. Here, we present a comprehensive study of closely related species of O. sativa complex and attempt to identify all reproductive barriers limiting hybridization. Understanding the molecular basis and the evolutionary forces that caused these barriers to evolve will increase our knowledge of reproductive isolation.

Domestication: selection by early humans
Domestication is the result of a selection process by early farmers that led to the increased adaptation of a plant to cultivation and utilization by humans (Gross and Oslen 2010). The domestication process involves the repeated selection for desirable traits, resulting in the responsible gene mutations becoming fixed in the populations. In some cases, these mutations may lead to divergence and acquiring of variable degrees of isolation from their wild ancestors (Milla et al. 2015). After divergence, the domesticated species became dependent on humans for their reproduction and geographical spread (Milla et al. 2015).
Rice domestication under early human selection led to the intense morphological and physiological variations from its wild ancestors (Table 1) (Sweeney and McCouch 2007;Asano et al. 2011). These alterations resulted in the creation of high-yielding, uniform-germinating and densely planting present-day cultivated rice varieties. The most striking impact of domestication, which differentiated wild and cultivar populations into different reproductive and ecological realms, was changing rice from an outcrossing to an inbreeding crop under human selection for uniform traits. The wild species O. rufipogon and O. barthii are outcrossing, while O. sativa and O. glaberrima are almost entirely self-pollinated.
Oryza sativa has a limited degree of outcrossing owing to the short style and stigma, short anthers, limited pollen viability and the brief period between flower opening and pollen release (Tripatti et al. 2011). However, the wild ancestor O. rufipogon have large stigma, and long anthers (Fig. 3). The shift to selfing is associated with changes in flower morphology, matting patterns and reproductive investment that can in turn affect the extent of hybridization and gene flow between populations (Wendt et al. 2002;Coyne and Orr 2004;Martin and Willis 2007;Sicard and Lenhard 2011;Wright et al. 2013). This suggests that different suites of genes and the corresponding positive mutations that accumulated in the domesticated genotype would be absent from the wild genotype. Therefore, when crosses are made between the domestic and wild populations, domestication-related loci may not interact well with each other.

Crop improvement: selection by breeders
Crop improvement involves the incorporation of as many desirable characteristics as possible into a single variety to make it a superior variety. Rice crop improvement tends to produce new varieties with increased yields, enhanced grain characteristics and nutritional values, and an increased resistance against various biotic and abiotic factors. Selective breeding is also associated with the spreading of genes among/or between populations (Chen et al. 2016). As an example, the domestication allele of the reduced shattering gene sh4 in rice originated in the japonica group and spread to the indica group through selective breeding (Gross and Olsen 2010). Selective breeding favours the white grains of domesticated varieties compared with the pigmented grains of wild rice. The gene OsC1 is responsible for the grain pigmentation (Saitoh et al. 2004). OsC1 is tightly linked with the sterility locus S5 (Saitoh et al. 2004). S5 is a major locus controlling embryo-sac fertility and is responsible for the low fertility observed in the indica-japonica hybridizations. The S5 locus contains three alleles: an indica allele (S5i), japonica allele (S5j) and a neutral allele (S5n) . S5i and S5j are each compatible with S5n, but the combination of S5i and S5j leads to hybrid sterility. Independent mutations to OsC1 leading to lighter grain colour occurred on the background of S5i and S5j, respectively. Selection for the light-colour OsC1 allele indirectly led to increased frequencies of the S5i and S5j alleles at the S5 locus. Thus, selective breeding for OsC1 results in the increase in frequency of S5 genes in the population and an increase in sterility barrier which indicates that S5 locus and post-zygotic isolation can arose as a by-product of genetic hitchhiking between tightly linked genes in rice (Saitoh et al. 2004). Another well-documented example results from the tight linkage between the yield establishment gene Gna1 and the sterility locus S35. Again, selection for the yield increasing gene might have indirectly favoured the retention of a Divergence between species is associated with a set of morphological, developmental and genetic changes which create reproductive barriers between them and prevent them from breeding. Hybridization between two diverged species can result in maladapted, non-viable or infertile hybrids. Barriers are listed in the order in which they occur.
However, the hybridization between cultivated and wild species is often limited by linkages between desirable and undesirable genes that may hinder the production of lines with desirable agronomic characteristics (Xiao et al. 1998). For example, many desirable agronomic traits in wild rice are often linked with the easy shattering phenotype (Xiao et al. 1998). This suggests that linkage  between favourable and unfavourable alleles may limit introgression.

Pre-zygotic reproductive barriers
In some well-studied cases of plants, pre-zygotic reproductive barriers are found to be stronger and lead to nearly complete reproductive isolation (Lowry et al. 2008). The adaptation of species or subspecies to different environments has been recognized as an important reproductive isolating barrier (Baack et al. 2015). Local adaptation involves the evolution of traits in response to different environmental conditions (Baack et al. 2015). These traits include abiotic stress tolerance, breeding times, changes in floral characteristics, flowering time and gametes compatibility that can create reproductive incompatibility between populations (Baack et al. 2015 (Zheng and Ge 2010;Liu et al. 2015). Incompatibility in pollen-pistil interactions provides a strong pre-zygotic reproductive barrier (Oka 1988;Harushima et al. 2002;Lowry et al. 2008

Post-zygotic reproductive barriers
Post-zygotic reproductive barriers preclude the development of viable fertile hybrid progeny. These barriers in rice begin at hybrid seed development and continue until the hybrid plant reaches the seed-producing stage (Tiffin et al. 2001). At each developmental stage in a plant's life cycle, a barrier prevents the hybrid progeny from reaching the next developmental stage. These post-zygotic barriers are further classified based on the stage of occurrence.

Hybrid inviability: an endosperm-based hybridization barrier
The endosperm is the basic nutrient source for the developing embryo (Lafon-Placette and Köhler 2016). Any abnormality during endosperm development ultimately leads to embryo arrest and seed failure (Lafon-Placette and Köhler 2016). In dicot species like Arabidopsis, the endosperm supports embryonic growth and disappears soon after cellularization (Bushell et al. 2003). In monocot species, like rice, the endosperm continues to proliferate and support seedling growth even after germination (Sekine et al. 2013). Seeds that lack properly developed endosperm fail to germinate. The endosperm is a highly sensitive tissue and requires a relative maternal to paternal genome dosage ratio of 2:1 for successful development and requires a highly specific balanced gene expression (Lafon-Placette and Köhler 2016). Defects in parental genome dosages, caused by either interspecific crosses or interploidy crosses, are the main reasons for hybrid endosperm defects (Ishikawa et al. 2011;Sekine et al. 2013). Genomic imprinting, whereby some genes are expressed in parental origin-specific manners, serves as the molecular basis for parental genome dosage effects (Matsubara et al. 2003;Ishikawa et al. 2011;Chen et al. 2016). Generally, outbreeding is thought to increase the intensity levels of the parental genome conflict and the associated genomic imprinting, while inbreeding is thought to reduce those (Lafon-Placette and Köhler 2016  (Chu and Oka 1970). Parental genome conflicts arising from ploidy variations have also been reported in a cross between diploid and tetraploid japonica rice (Sekine et al. 2013). Endosperm-based hybridization barriers are the evolutionary forces associated with parental genome conflict, which might arose as a result of a shift in mating processes (i.e. from outbreeding to inbreeding) and thus established a barrier to interploidy and interspecific hybridization in rice.

Hybrid weakness: a post-embryonic-stage hybridization barrier
Hybrid weakness also known as hybrid necrosis is a post-embryonic-stage barrier frequently observed in plant taxa (Bomblies and Weigel 2007). The phenotype of hybrid weakness is similar to the symptoms associated with disease responses (Bomblies and Weigel 2007). Hybrid weakness has been reported in many crosses involving O. rufipogon × O. sativa, O. barthii × O. glaberrima as well as between the subspecies of O. sativa (indica × indica and japonica × japonica hybridizations) (Ichitani et al. 2011;Zhang et al. 2012;Chen et al. 2013Chen et al. , 2014. Early embryogenesis is normal, and normal seedlings are established . However, at later stages, the seedlings fail to grow properly. The hybrid seedlings have retarded growth rates, with a pale yellow phenotype, and they undergo wilting and necrosis (Chen et al. , 2014. Although, many cases of hybrid weakness have been reported in rice, the exact molecular mechanism underlying hybrid weakness is still not well known. Previous studies have suggested that hybrid weakness is under the control of complementary gene either dominant or recessive (Amemiya and Akemine 1963;Chu and Oka 1970;Ichitani et al. 2007Ichitani et al. , 2011Saito et al. 2007;Kuboyama et al. 2009;Chen et al. 2014). Some studies have suggested that hybrid weakness is due to the activity of defence-related genes which was further confirmed by the high activity of F 1 hybrids against pathogens (Bombiles and Weigel 2007; Chen et al. 2014). Recently, genes conferring hybrid weakness in O. rufipogon × O. sativa F 1 hybrids have been cloned (Chen et al. 2014). Molecular studies identifying the casual gene of hybrid weakness will help in detailed understanding of the phenomena.

Hybrid sterility: a reproductive-stage hybridization barrier
Hybrid sterility is the most common form of postzygotic hybridization barrier in plants (Ouyang et al. 2010). A well-documented example is the hybrid sterility observed in crosses between O. sativa and O. meridionalis, O. sativa and O. glaberrima and between indica and japonica subspecies of O. sativa (Ouyang et al. 2010;Yu et al. 2018). The hybrids obtained are robust in their vegetative growth; however, the progeny are often sterile and cannot produce the next generation. The physiological determinants of reproductive failure in rice include female gamete abortion, pollen sterility, reduced affinity between the uniting gametes, panicle growth rate and ovary growth Long et al. 2008;Mizuta et al. 2010). Approximately 50 loci have been identified as being involved in the control of indica-japonica hybrid sterility Long et al. 2008;Yamagata et al. 2010;Yang et al. 2012a;Kubo et al. 2016;Li et al. 2017). Several sterility loci have been identified and mapped as a single genetic loci in O. glaberrima (Hu et al. 2006;Li et al. 2011). Further studies relating to gene characterization will help in understanding the molecular mechanism underlying hybrid sterility. Recently, Yu et al. (2018) mapped a sterility locus that contains two tightly linked open reading frames (ORFs) that confers hybrid sterility in F 1 hybrids derived from crossing O. sativa × O. meridionalis. One of the ORFs encodes a toxin, which affects the development of pollen, and the other encodes an antidote, which is required for pollen viability. Hybrid breakdown is the weakness and sporophytic sterility found in the F 2 and advanced generations, and can be genetically different from F 1 weakness or sterility (Okuno and Fukuoka 1999). Hybrid breakdown has been detected in many crosses of rice (Okuno and Fukuoka 1999).

Genetic Models for the Evolution of Reproductive Barriers in Rice
Three genetic models have been developed to explain the kinds of genetic changes that occur to cause reproductive isolation in rice.

Complementary epistasis interaction between two loci
The simplest genetic explanation of the complementary epistatic interactions between two loci that lead to hybrid inferiority was proposed by Bateson, Dobzhansky and Muller and termed the BDM model (Coyne and Orr 2004). These two loci may be duplicate gene copies or the same locus evolving repeatedly in one species or differently in two species. Figure 4A show a model representation of two-locus interaction causing hybrid incompatibility. Reviewing the literature of hybrid incompatibility cases in rice, a two-locus interaction seems to be the most common cause. A two-locus interaction leading to hybrid weakness was recently reported in O. rufipogon and O. sativa (indica) hybridization (Chen et al. 2014). The hybrids were found to have elevated immune responses similar to previously reported in Arabidopsis (Bombiles and Weigel 2007). The genes in hybrid weakness were found to encode defence-related proteins against pathogens and that incompatible allelic combinations appear to induce autoimmune responses (Chen et al. 2014). Similar two-locus interaction between pathogen resistance genes and their interacting partners has been implicated in hybrid weakness between the indica and japonica subspecies of cultivated rice ).

Differential silencing
Molecular divergence owing to the divergent evolution of duplicated genes has been identified as a potential source of reproductive isolation (Werth and Windham 1991). After the duplication event, it is highly probable that both or one of the genes will mutate (Prince and Pickett 2002). Therefore, it can be assumed that an independent mutational event causes a pair of paralogous genes to undergo divergent evolutionary paths and becomes fixed in the two diverging populations. When the two populations hybridize, the populations carrying either of the functional copies develop properly, whereas the hybrids receiving the two silenced copies will have reduced fitness levels. The loss of the duplicated gene that encodes an essential protein for pollen development causes pollen sterility in the F 1 interspecific hybrids of O. sativa and O. glumaepatula (Yamagata et al. 2010). Additionally, the independent disruption of duplicated genes DOPPELGANGER1 and DOPPELGANGER2 in the indica and japonica cultivars, respectively, causes pollen sterility in inter-subspecific F 1 hybrids (Fig. 4B) (Mizuta et al. 2010). Likewise, duplication and the reciprocal silencing due to lossof-function mutation of duplicate genes S27 and S28, encoding a mitochondrial protein, were observed in the F 1 interspecific hybrids of O. sativa and O. glumaepatula rice. An ancestral population splits into isolated populations that diverge genetically as a result of fixation of an independent mutation at each locus. If the two diverged populations hybridize, 1 in 4 F 1 hybrids and 1 in 16 F 2 progeny will have incompatible genotypes. (B) Gene duplication in post-zygotic reproductive isolation in rice. An ancestral population undergoes a duplication event followed by divergence due to a mutation (either a gain of deleterious function or loss of function). In the F 1 hybrid, these two mutated alleles are incompatible (Mizuta et al. 2010). (C) A simple illustration for three tightly linked genes at the S5 locus. K1 and K2 represent Killer genes, and P represent the Protector gene. The haplotype K1K2P represents a balance between killing and protecting of the gametes according to the genetic model in the S5 system. Indica and japonica rice have independent mutations in these three linked genes, with japonica haplotype being k 1 K2p and indica haplotype being K1k 2 P. The gametes carrying loss of protector (p) will be non-viable. (Yamagata et al. 2010). Thus, gene duplication and subsequent mutations may play important roles in establishing reproductive barriers in rice.

Genic interactions
A well-documented example of genic interactions leading to hybrid sterility comes from the tight linkage of three genes at the S5 locus causing female gamete abortion and hybrid sterility in indica japonica hybridization (Yang et al. 2012b). Two of these three genes constitute the killer system which will preferentially kill the gametes which lack the protector. The divergence in any of these three genes does not occur independently and the evolution of one gene is expected to be conditional on the evolution of another gene. Therefore, a non-functional mutation in the killer would not cause the loss-of-function mutation in the protector. Rather, once the loss-of-function mutation in the killer exists, loss-of-function mutation in the protector would no longer be deleterious and might drift to fixation. Subsequent evolution in the indica and japonica subspecies has silenced different parts of the killer/protector system and neither subspecies has a functional killer phenotype. In heterozygotes, different genotypes will be formed of these three-gene combinations, hybrid sterility will appear owing to the deleterious interaction between the killer loci without protection by the protector (Yang et al. 2012) (Fig. 4C). Similar findings have been reported by Yu et al. (2018) where a selfish genetic element that encodes for a toxin and antidote that affects pollen viability causes hybrid sterility in O. meridionalis and O. sativa hybrids.

Crop-to-weed gene flow
Weedy rice is one of the troublesome weeds that grows sympatrically with the cultivated rice in rice fields (Nadir et al. 2017). Variable level of gene flow has been recorded to occur between cultivated rice and weedy rice (Langevin et al. 1990;Gealy et al. 2003;Nadir et al. 2017). One study has reported that these two sympatric populations can hybridize with each other freely without any sterility issue which indicates that fewer postzygotic barriers exist between them (Craig et al. 2014). Furthermore, gene flow in reverse order from weed to crop has also been reported (Burgos et al. 2008;Serrat et al. 2013). Gene flow in either direction may have detrimental environmental consequence such as the evolution of undesired agronomic characters in rice crop or the evolution of more aggressive weeds that are difficult to control.

Crop-to-wild gene flow
Domesticated rice mate with wild relatives and variable level of gene flow has been reported to occur (Langevin et al. 1990;Majumder et al. 1997;Song et al. 2002). Oryza sativa frequently hybridize with O. rufipogon. The morphological differences in floral traits suggest that gene flow will predominantly be from O. sativa to O. rufipogon and the gene flow shows distinct geographic patterns and varies with the O. sativa subspecies . The gene flow from domestic to wild rice has the potential to affect the genetic structure of wild rice . Wild-to-crop gene flow can be expected even though the frequency is lower than that of crop-to-wild gene flow. The asymmetric gene flow recorded in the O. rufipogon and O. sativa should be further explored to gain an insight into the forces and mechanism that determine reproductive isolation. These data can also help us in understanding the patterns of population-specific reproductive isolation in rice.

Crop-to-crop gene flow
The two domestic rice species O. sativa and O. glaberrima represent parallel domestications from different progenitors on two different continents assisted by nearly same set of genes (Purugganan 2014). This suggests a high level of homology between the genomes of these Oryza species and the possibility to carry out gene transfer between them but this is difficult because of strong reproductive barriers between them (Chu and Oka et al. 1969). Similarly, gene exchange between indica and japonica subspecies of O. sativa has also been reported (Yang et al. 2012b). Gene exchange between these rice subspecies would be highly beneficial to rice breeding practices, but the sterility barriers hinder the exchange .
In this condition, we may assume that a certain rate of gene exchange takes place across the complex of isolating barriers (Zheng and Ge 2010). Perhaps, a barrier protects populations from reproductive waste, whereas introgression across the barrier offers genetic variability to the populations, giving rise to a balance between isolation and hybridization (Chu et al. 1969). With the rapid advances in transgenic biotechnology, several transgenic crop varieties have been developed and released into markets (Kwit et al. 2011). Synthetic genotypes with desired properties have been developed in rice (Croughan 1998;Paine et al. 2005). Crop-to-weed or crop-to-wild gene flow is of special concern when transgenes are involved (Lu and Snow 2005). These transgenes can escape and introgress into the wild and weedy population with potentially serious consequences (Lu and Snow 2005). The development of genetically modified crops has been accompanied by efforts to development of intrinsic genetic barriers to prevent transgene flow between synthetic and wild/weedy populations. These include interfering with pollination and fertilization using maternal inheritance and male sterility, terminating transgenic fruit/seed development (seed sterility), selectively terminable transgenic lines or compromising the fitness of hybrids that have acquired positive survival traits from crop genes through introgression (transgenic mitigation) (Suh 2008;Kwit et al. 2011;Nadir et al. 2017). This may also have the side effect of constructing intrinsic barriers to gene flow between independently synthesized transgenic lines that originate from the same ancestral forms (Schluter and Pennel 2017).

Future Research Perspective
Rice serves as an excellent model for understanding speciation, and some species (O. rufipogon and O. nivara) are in the early stages of divergence. Therefore, a comprehensive understanding of all the forces and mechanisms that drive reproductive isolation is required. Some additional studies needed in this field are listed below: 1. Analysis of the quantification of the strength of the individual barrier and its contribution to the reproductive isolation in rice would allow us to better evaluate the importance of gene flow and reproductive isolation. 2. Although species of different ploidy are generally reproductively isolated from each other, gene flow across ploidy barriers has been reported in many plant taxa (Chapman and Abbott 2010;Pinheiro et al. 2010). However, in rice introgression across ploidy barrier is not well studied. Understanding the gene flow in ploidy variation and its impact on the morphology and ecology of rice species will be helpful in understanding the diversification and evolution of rice. 3. In addition to the crop-wild (interspecific) and indicajaponica (inter-subspecific) reproductive barriers, some barriers also exist between intra-subspecific hybridizations (indica-indica and japonica-japonica) (Zhang 2012;Fu et al. 2013). The hybrid weakness observed in the intra-subspecific hybridization (indica-indica and japonica-japonica) needs evolutionary as well as molecular analysis. 4. Few genes involved in reproductive isolation in rice have been identified and cloned Long et al. 2008;Chen et al. 2013Chen et al. , 2014Yu et al. 2018). However, the biochemical and molecular mechanisms, and the relationships between the causal genes are still largely unknown. For a complete understanding of these processes, further efforts are needed to characterize the mechanisms that control how gene products function to induce hybrid dysfunction at the molecular, cellular and organ levels. These studies will not only improve our understanding of reproductive isolation but also help to improve crop breeding strategies.